Asset 20 8 2
Does AI recommend your business? Run the free check →

Join 15,000 business owners, marketers and entrepreneurs. The Sunday newsletter you'll be annoyed only arrives once a week.

Article

Claude Prompts for LinkedIn Posts That Sound Like You

If you are skim reading
The short version: Claude writes better LinkedIn posts than ChatGPT in my experience, but only if you feed it your actual sentences, not a vague topic.

The short version: Claude writes better LinkedIn posts than ChatGPT in my experience, but only if you feed it your actual sentences, not a vague topic. Give it raw material (voice notes, half-finished thoughts, old posts that did well) and a strict structure, and it stops sounding like every other "thought leader" post clogging your feed. Give it a blank prompt like "write me a LinkedIn post about leadership" and you'll get the same beige nonsense everyone else is pumping out.

Why most Claude LinkedIn prompts produce rubbish

I'll say the uncomfortable bit first. The reason your AI-written LinkedIn posts sound hollow isn't the AI. It's that you asked it to invent an opinion you don't hold. Claude is very good at structure, rhythm and tightening a sentence. It is not good at having lived your week. When you type "write a post about the importance of networking" you're asking a model with no experience of your Tuesday to manufacture one, and it shows in every generic LinkedIn post with three bullet points and a question at the end.

The fix isn't a cleverer prompt. It's feeding Claude something real to work with first, then using the prompt to shape it.

The three-part prompt structure that works

Every decent Claude prompt for LinkedIn has three parts: raw material, a structural instruction, and a voice constraint. Skip any one of them and you get slop.

  • Raw material: a paragraph of your actual thinking, a voice note transcript, a client email you can anonymise, or a story from your week.
  • Structural instruction: hook line, then 3 to 5 short paragraphs, then one closing line. Tell Claude exactly how many lines before the first line break, because LinkedIn truncates after about three lines on mobile.
  • Voice constraint: paste in two or three of your old posts and say "match this rhythm and vocabulary, not generic LinkedIn tone."

The prompt I'd use

Here's a template worth copying exactly:

"I'm going to give you a messy paragraph of my thinking and two examples of my past LinkedIn posts. Write a new LinkedIn post using the messy paragraph as the core idea. Match the sentence length, tone and punctuation style of my examples, not standard LinkedIn style. Structure: a one-line hook that works even if the post gets cut off on mobile, then 4 short paragraphs of 1 to 3 sentences each, then a single closing line with no question mark unless it's a real question I'd ask a friend. No emojis. No bullet point lists unless the content needs one. Do not use the words open up, journey, game-changer or delve."

That last line matters more than people think. Claude, like every model, defaults to a specific vocabulary when it's uncertain. Naming the words you want banned does more work than any amount of "write in a human tone" instructions.

A worked example

Say you run a small bookkeeping firm in Manchester with four staff, and you want to post about a mistake a lot of small business owners make with cash flow. Instead of prompting Claude with "write a LinkedIn post about cash flow mistakes," you'd first write a rough paragraph like this and paste it in as raw material:

"Loads of small business owners think they're profitable because their bank balance looks healthy in December. But that's often just because a big invoice landed early and three supplier payments haven't gone out yet. I see this every January when people panic because the cash has vanished and they assumed it was theirs to spend."

Then you'd use the three-part prompt above. What comes back from Claude, in my experience with this kind of input, is something that opens with a line like "Your bank balance is lying to you right now" rather than "In today's business landscape, cash flow management is more important than ever." That gap, between a specific lie and a vague landscape, is the entire difference between a post people stop scrolling for and one they skim past.

From there you'd edit two or three lines yourself, because Claude still occasionally reaches for a tidy conclusion that sounds more like a LinkedIn influencer than you. That final pass is non-negotiable, which I'll come back to below.

Prompts for five common LinkedIn post types

You don't need one magic prompt. You need slightly different instructions depending on what the post is trying to do.

  • The opinion post: "I'm going to state an opinion in one sentence. Write a LinkedIn post that defends this opinion using one concrete example from my industry. Don't hedge the opinion in the first line. Don't soften the ending with 'what do you think.'"
  • The lesson-from-a-mistake post: "Turn this into a LinkedIn post structured as: the mistake, what it cost (time, money or a client), what I'd do differently. Keep the cost specific, a number or a timeframe, not 'it was costly.'"
  • The listicle post: "Turn these five rough points into a numbered LinkedIn post. Each point gets one sentence of explanation maximum. Cut anything that sounds like generic advice found in every other post on this topic."
  • The behind-the-scenes post: "Write this as if I'm telling a colleague over coffee, not presenting to a conference. Short sentences. No 'I'm thrilled to announce.'"
  • The data or stat post: "Here's a stat and my paragraph of reaction to it. Lead with the stat in the first line, then my reaction, then one practical takeaway a reader could apply this week."

If you're building a weekly content rhythm around these, it's worth pairing this approach with the Claude prompts for social media captions I've written up separately, since the voice-matching principle carries across platforms even though the structure for LinkedIn is different from Instagram or X.

The uncomfortable bit about editing

I'll be blunt about something most posts on this topic avoid saying out loud: if you post what Claude gives you on the first draft, people who know you will clock it within two sentences. Not because the writing is bad, but because it's too even. Your real writing has a slightly odd rhythm, a sentence that runs long, a word you overuse. Claude's first draft smooths all of that out, which makes it read as competent and impersonal at the same time.

Work with me

Want AI doing the heavy lifting in your marketing?

I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.

The step that fixes this takes about 90 seconds and almost nobody does it: read the draft out loud, and anywhere it doesn't sound like something you'd say to a person's face, change it. That's it. Not a rewrite, a read-through with a pen. I'd treat this the same way I treat any AI-assisted editing, which is covered in Claude prompts for editing your own writing, because the skill of catching "AI voice" in your own drafts is the single most valuable editing skill right now and it's barely taught anywhere.

Feeding Claude better raw material in the first place

The quality of your LinkedIn posts is capped by the quality of what you hand Claude, not by the cleverness of the prompt. A few ways to get better raw material without much effort:

  • Record a 60-second voice note after a client call or a frustrating meeting and get it transcribed. Paste the transcript in as the raw material instead of typing a summary. Transcripts keep your actual phrasing, summaries lose it.
  • Keep a running note of one-line observations from your week. "Client asked me to fix their website before fixing their offer" is more useful to Claude than "today I learned about priorities."
  • Screenshot or paste three of your best-performing old posts (by comments, not just likes) and ask Claude to describe the pattern in them before you ask it to write anything new. This gives you a style brief you can reuse across dozens of future prompts.

If you're planning content further ahead than one post at a time, the structured thinking in the RICE framework for AI prompts is worth applying here too, since a vague brief produces a vague post regardless of which model you're using.

Where this goes wrong at scale

I've seen the pattern enough times to name it: someone gets one great Claude-assisted post, gets 40 comments on it, and then tries to replicate it by running the same prompt with a new topic every day for a month. The posts start reading identically. Same hook structure, same closing line shape, same rhythm. LinkedIn's own audience, and the algorithm, notice repetition faster than you'd think. Rotate your prompt structures across the five types above, and change your voice-sample posts every few weeks so Claude isn't locked onto one mood from three months ago.

If LinkedIn is one piece of a wider content operation, it's also worth getting the proofreading pass separated from the drafting pass entirely. I keep those as two distinct prompts rather than one, which is the approach laid out in Claude prompts for proofreading without losing your voice. A draft prompt chasing structure and a proofread prompt chasing your voice back are doing two different jobs, and asking one prompt to do both tends to flatten the result.

A note on length and posting rhythm

LinkedIn's algorithm still rewards posts that keep people on the platform, which is why a 150 to 300 word post with short line breaks tends to outperform a dense 600-word essay, in my experience. Ask Claude for a word count in the prompt itself: "keep this under 220 words, short paragraphs, no paragraph longer than three lines." That one instruction solves more formatting headaches than any amount of fiddling after the fact.

Frequently asked questions

Is Claude better than ChatGPT for LinkedIn posts?

In my experience Claude tends to produce tighter sentences and fewer stock phrases out of the box, but the gap closes almost completely once you give either model strong raw material and a clear structural prompt. The model matters less than what you feed it.

How do I stop my Claude-written LinkedIn posts sounding like AI?

Paste in two or three of your own old posts as a voice reference, name specific words you want banned, and read the final draft out loud before posting, changing anything that doesn't sound like something you'd say to a person's face.

What's the ideal length for a LinkedIn post written with Claude?

Aim for 150 to 300 words with short paragraphs, no more than three lines per paragraph, since LinkedIn truncates posts on mobile after about three lines and short paragraphs keep people tapping "see more."

Can I use the same Claude prompt for every LinkedIn post?

No, and you shouldn't try. Rotate between opinion, lesson, listicle, behind-the-scenes and data-led prompt structures, otherwise your posts start reading identically and both your audience and the algorithm will notice the repetition.

Run a business, a brand or a blog? I feature tools, products and expert guest posts for an audience of 15,000. See how it works

Published and maintained by the Lilach Bullock team, covering marketing, AI and business growth.
Your buyers are asking AI who to use. Does it say you?

See for free whether ChatGPT, Claude, Perplexity, Gemini and Google name you, and get the plan to become the answer.

Check my AI visibility →
Sundays only

Get the Sunday newsletter.

One email a week. AI experiments, marketing tactics, and the workflows Lilach is building right now in her own business.

Subscribe free

Let’s get your marketing running on AI.

Book a free 30-minute call

We figure out what you need, where AI fits in, and what working together would look like.

Book the call →

Or take the 30-second calculator

You’ll see the hours and the money quietly leaking out of your week, and the three workflows worth building first.

Take the calculator →

Or grab the free AI resource library

Prompt packs, templates, checklists, and swipe files. The exact tools I build for paying clients. Yours, free.

Get the library →
Keep reading

More from the blog.